Suhail(@Suhail)

There is going to be 10x more distillation research with the threat of open weight model restriction...

Suhail (@Suhail) — Thu, 23 Jul 2026 12:37:56 GMT

There is going to be 10x more distillation research with the threat of open weight model restrictions. It will Streisand effect the whole community into making it even more data / compute efficient. Just to troll the government. Silly game to play.

💬13🔄2❤️65👀5229📊18

⚡ Powered by xgo.ing

Correct. You want safe models? This is what you do. Tons of software we use everyday is made by peo...

Suhail (@Suhail) — Wed, 22 Jul 2026 23:17:02 GMT

Correct. You want safe models? This is what you do.

Tons of software we use everyday is made by people of all nationalities. When it's indispensable for American National Security, we invest in locking it down. You do not need to ban technology that can be made safe.

Parker Conrad@parkerconrad

A set of open weights has no nationality. A model hosted on American infra and controlled by an American company is as American as apple pie.

🔗 View Quoted Tweet

💬11🔄10❤️111👀12483📊17

⚡ Powered by xgo.ing

Glad to see Elon's position on this. I really think highly of him coming out on it even though it do...

Suhail (@Suhail) — Wed, 22 Jul 2026 18:24:18 GMT

Glad to see Elon's position on this. I really think highly of him coming out on it even though it doesn't directly benefit him. These models also create competition for him. He could easily stay silent and let it happen.

💬23🔄21❤️403👀32635📊63

⚡ Powered by xgo.ing

It sounds like lobbying is going well!

Suhail (@Suhail) — Wed, 22 Jul 2026 17:57:31 GMT

It sounds like lobbying is going well!

Suhail@Suhail

The main thing I've learned this week is that Anthropic's lobbyists are world-class. They are absolutely killing it at a level this website cannot begin to comprehend. Level 0 is posting on X. These guys are Jedi master level playing the game.

🔗 View Quoted Tweet

💬4🔄20❤️227👀21490📊24

⚡ Powered by xgo.ing

Now I can go to bed. Good luck little AI research agent. May your wandb metrics avoid collapse.

Suhail (@Suhail) — Wed, 22 Jul 2026 06:05:52 GMT

Now I can go to bed. Good luck little AI research agent. May your wandb metrics avoid collapse.

💬6🔄2❤️48👀13911📊10

⚡ Powered by xgo.ing

If you want to help adjust the narrative / direction of open weight models, use of Chinese OSS, avoi...

Suhail (@Suhail) — Tue, 21 Jul 2026 17:29:55 GMT

If you want to help adjust the narrative / direction of open weight models, use of Chinese OSS, avoidance of reg capture, etc. please DM me. 

We need credible AI researchers to help and I am working with people connected to this situation brewing.

💬1🔄0❤️27👀2695📊4

⚡ Powered by xgo.ing

Anthropic has figured out the main game is making everyone operate in a state of fear. It's a consta...

Suhail (@Suhail) — Tue, 21 Jul 2026 17:09:01 GMT

Anthropic has figured out the main game is making everyone operate in a state of fear. It's a constant source of PR. The government optimizes for minimizing downside risk, not upside.

Internal government IT systems insecure? Chinese lead in cyber warfare? Chinese IP theft at the expense of American co? Capture of American data? Check. Check. Check. Check. Say it all, see what sticks.

How do you thread the needle between those things, American regulatory capture, and ensuring your revenue machine? Easy!

Ban the open model weights we can make safe, can operate within American, and create competition for Anthropic. Even the government *wants* competition to avoid Anthropic's policy / monopolization of intelligence.

Do this instead of limiting usage of Chinese API based services which would be a more *nuanced* policy. Don't explain the difference between model weights, APIs, apps (like TikTok) to policy makers who aren't as savvy - just ban it if it has anything to do China.

Suhail@Suhail

The main thing I've learned this week is that Anthropic's lobbyists are world-class. They are absolutely killing it at a level this website cannot begin to comprehend. Level 0 is posting on X. These guys are Jedi master level playing the game.

🔗 View Quoted Tweet

💬38🔄46❤️442👀48719📊74

⚡ Powered by xgo.ing

The main thing I've learned this week is that Anthropic's lobbyists are world-class. They are absolu...

Suhail (@Suhail) — Tue, 21 Jul 2026 16:43:11 GMT

The main thing I've learned this week is that Anthropic's lobbyists are world-class. They are absolutely killing it at a level this website cannot begin to comprehend. Level 0 is posting on X. These guys are Jedi master level playing the game.

💬30🔄12❤️468👀127348📊83

⚡ Powered by xgo.ing

If Chinese AI gets added to the entity list it's going to be hard for any American corporation to us...

Suhail (@Suhail) — Mon, 20 Jul 2026 16:12:01 GMT

If Chinese AI gets added to the entity list it's going to be hard for any American corporation to use anything remotely tied to these openweight models. This means total market capture by the few American labs.

If you own intelligence, you can tax humanity.

💬41🔄14❤️193👀19670📊60

⚡ Powered by xgo.ing

It brings me tremendous peace to know Emil is completely aware of the regulatory capture going on he...

Suhail (@Suhail) — Sun, 19 Jul 2026 21:08:29 GMT

It brings me tremendous peace to know Emil is completely aware of the regulatory capture going on here. 💪

Going back to enjoying my weekend. Thank god.

Under Secretary of War Emil Michael@USWREMichael

So let me get this straight @deanwball This was about ensuring that DoD IT networks were not connected to adversary networks. You think your deep state regulatory capture scheme is equivalent to making sure our warfighter networks are not compromised? I take it back: 50 IQ difference (75 versus 125). Take a breather and enjoy the match.

🔗 View Quoted Tweet

💬9🔄12❤️337👀69095📊38

⚡ Powered by xgo.ing

They distilled humanity. The hypocrisy is unreal. The best gift is to give it away, at the lowest ...

Suhail (@Suhail) — Sun, 19 Jul 2026 00:34:36 GMT

They *distilled* humanity. The hypocrisy is unreal. The best gift is to give it away, at the lowest marginal costs, to let us prosper as a species. If intelligence casts a light cone, it is a moral imperative to extend it as diffusely as possible.

💬28🔄102❤️1171👀86113📊163

⚡ Powered by xgo.ing

The Chinese do not need to distill anymore. They've climbed up the proverbially tech tree that to cl...

Suhail (@Suhail) — Sun, 19 Jul 2026 00:21:37 GMT

The Chinese do not need to distill anymore. They've climbed up the proverbially tech tree that to climb higher, they need to do actual frontier work. Distillation is marginal.

Any lobbying or legislation that calls for banning open weight models in the name of "distillation" is total BS. This is a fight against future American innovation.

The labs know that open weight models completely commoditize their revenue generator. They know that we will optimize it to 1/100th the costs. They know that we will stop concentrating our workloads on them. If we start using open models, they lose the understanding of workloads that they can build products around to *replace* your products as they move up to the app layer.

Remember: *they* trained on humanity's data and didn't pay a cent. If you own intelligence, you can tax humanity.

Do not let them fool you.

💬127🔄353❤️2859👀174313📊501

⚡ Powered by xgo.ing

Product quality always seems to go down hill when you start thinking to yourself “You know, we sure ...

Suhail (@Suhail) — Fri, 17 Jul 2026 22:56:45 GMT

Product quality always seems to go down hill when you start thinking to yourself “You know, we sure need some moats. Can we think of some network effects to add this to thing?”

You often cannot add these things retroactively when they’re not inherent to the product.

💬13🔄7❤️181👀24296📊34

⚡ Powered by xgo.ing

Literally frog boiling the app layer. There needs to be a neutral party in this.

Suhail (@Suhail) — Fri, 17 Jul 2026 19:03:55 GMT

Literally frog boiling the app layer. There needs to be a neutral party in this.

💬1🔄0❤️33👀5633📊4

⚡ Powered by xgo.ing

A company will announce a blog post and some money and people will be like "Ok, but how do you compe...

Suhail (@Suhail) — Fri, 17 Jul 2026 17:55:45 GMT

A company will announce a blog post and some money and people will be like "Ok, but how do you compete with X company?" Homie, that company hasn't done anything.

💬4🔄2❤️10👀2181📊5

⚡ Powered by xgo.ing

If your core technology continues to get commoditized (i.e. Kimi 3), you must go up the stack. The l...

Suhail (@Suhail) — Fri, 17 Jul 2026 15:31:31 GMT

If your core technology continues to get commoditized (i.e. Kimi 3), you must go up the stack. The labs will continue to encroach on everyone in the app layer. You are their competition now.  

I would be very careful to concentrate on them and send them your trajectories! Figma -> Claude Design is just the beginning--expect 20+ of these.

How your users navigate and use your products is an invaluable dataset. With the advent of Fable 5 not practicing ZDR, you should be super aware of this.

💬68🔄53❤️846👀76273📊187

⚡ Powered by xgo.ing

It's pretty cool that this technology that you have to have $10B to make is just freely available to...

Suhail (@Suhail) — Fri, 17 Jul 2026 15:25:20 GMT

It's pretty cool that this technology that you have to have $10B to make is just freely available to do what you want with and will get optimized.

OTOH, you must realize the trade off this too: the labs will encroach on everyone in the app layer. You are their competition now.

💬0🔄0❤️1👀379

⚡ Powered by xgo.ing

I see a lot of “without a doubt, they do” and not a lot of evidence or prove of life on methods. You...

Suhail (@Suhail) — Fri, 17 Jul 2026 14:37:30 GMT

I see a lot of “without a doubt, they do” and not a lot of evidence or prove of life on methods. You cannot just SFT on Fable 5 traces and expect a nearby equivalent to pop out of the oven. You can climb up a little bit but things like OPSD, for example, don’t work yet.

💬0🔄3❤️23👀3122📊3

⚡ Powered by xgo.ing

Well deserved. Valar is going to be huge.

Suhail (@Suhail) — Fri, 17 Jul 2026 03:12:58 GMT

Well deserved. Valar is going to be huge.

julia hornstein@julia_hornstein

nuclear startup @valaratomics is in talks to raise $1 b at a $6 b valuation including the investment, w @sequoia in talks to lead

🔗 View Quoted Tweet

💬6🔄1❤️50👀22148📊8

⚡ Powered by xgo.ing

Talked to some great friends today and completely lost track of time. I have a rule these days where...

Suhail (@Suhail) — Fri, 17 Jul 2026 00:45:08 GMT

Talked to some great friends today and completely lost track of time. I have a rule these days where if a conversation feels very meaningful with a friend, I will not look at the time/schedule and just let it ride. Has led to the best unregretted time.

💬8🔄6❤️180👀16088📊26

⚡ Powered by xgo.ing

Every single credible researcher I've talked to these past few weeks has said the "distillation" fro...

Suhail (@Suhail) — Thu, 16 Jul 2026 16:40:12 GMT

Every single credible researcher I've talked to these past few weeks has said the "distillation" from Chinese labs is way overexaggerated. Have you tried distilling without logits yourself? Show me the results. 

Narrative violation: maybe the Chinese are actually good.

💬101🔄85❤️1643👀115479📊265

⚡ Powered by xgo.ing

Interestingly, you can almost predict what's coming next based on the timeline. As my son would say:...

Suhail (@Suhail) — Thu, 16 Jul 2026 16:18:40 GMT

Interestingly, you can almost predict what's coming next based on the timeline. As my son would say: "Oh, there's a pattern!"

Hint: 🐋

💬3🔄1❤️32👀5371📊6

⚡ Powered by xgo.ing

I was curious about the last 4 of months of AI releases: Opus 4.7 · 4/16 Kimi K2.6 · 4/20 GPT-5.5 ·...

Suhail (@Suhail) — Thu, 16 Jul 2026 16:11:45 GMT

I was curious about the last 4 of months of AI releases:

Opus 4.7 · 4/16
Kimi K2.6 · 4/20
GPT-5.5 · 4/23
DeepSeek V4 Pro · 4/24
Opus 4.8 · 5/28
Fable 5 · 6/9
Kimi K2.7 · 6/12
GLM-5.2 · 6/16
Grok 4.5 · 7/8
GPT-5.6 · 7/9
Muse Spark 1.1 · 7/9
Inkling · 7/15
Kimi K3 · 7/16

💬32🔄33❤️589👀61918📊104

⚡ Powered by xgo.ing

The ability for AI to visualize things for me has been very helpful. Previously it'd be far too much...

Suhail (@Suhail) — Thu, 16 Jul 2026 15:43:22 GMT

The ability for AI to visualize things for me has been very helpful. Previously it'd be far too much work to build an application to inspect all kinds of data but now it's so easy and it helps you really understand the data you're working with when you train these models.

💬10🔄1❤️66👀8345📊18

⚡ Powered by xgo.ing

Jeez, who isn't going into the make-RL-env business???

Suhail (@Suhail) — Thu, 16 Jul 2026 05:46:46 GMT

Jeez, who isn't going into the make-RL-env business???

💬18🔄3❤️163👀34671📊39

⚡ Powered by xgo.ing

The vertical integration of post-training infra as a service has truly begun.

Suhail (@Suhail) — Wed, 15 Jul 2026 18:28:59 GMT

The vertical integration of post-training infra as a service has truly begun.

Thinking Machines@thinkymachines

Today, we are introducing Inkling.

Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.

thinkingmachines.ai/news/introduci…

Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵

🔗 View Quoted Tweet

💬2🔄9❤️237👀36649📊27

⚡ Powered by xgo.ing

If you sell synthetic data, real user prompts that people use in AI tools, analyzed user trajectorie...

Suhail (@Suhail) — Wed, 15 Jul 2026 16:26:44 GMT

If you sell synthetic data, real user prompts that people use in AI tools, analyzed user trajectories in your own products, etc please DM. Have some questions!

💬8🔄2❤️32👀9172📊11

⚡ Powered by xgo.ing

Must relocate to SF.

Suhail (@Suhail) — Wed, 15 Jul 2026 16:21:37 GMT

Must relocate to SF.

💬1🔄0❤️2👀1562📊1

⚡ Powered by xgo.ing

I just realized we passed the 1 year anniversary of Soham Parekh incident. Things have changed so mu...

Suhail (@Suhail) — Wed, 15 Jul 2026 00:48:30 GMT

I just realized we passed the 1 year anniversary of Soham Parekh incident. Things have changed so much that possibly someone could do that many jobs outside of tech. 

He emailed me a little while ago.

💬28🔄7❤️798👀131309📊101

⚡ Powered by xgo.ing

Perplexity is very good at building these benchmarks. Whenever I study them in-depth, it’s always th...

Suhail (@Suhail) — Wed, 15 Jul 2026 00:21:08 GMT

Perplexity is very good at building these benchmarks. Whenever I study them in-depth, it’s always the strongest in the field. DRACO was excellent compared to all the other Deep Research ones, for instance.

Perplexity@perplexity_ai

We’re open sourcing WANDR.

WANDR is an internal benchmark we built and used for building deep and wide research capabilities inside Perplexity Computer.

research.perplexity.ai/articles/wandr…

🔗 View Quoted Tweet

💬16🔄10❤️148👀69049📊36

⚡ Powered by xgo.ing

11/ banging my head against the proverbial hill climbing wall but met a bunch of people who set me i...

Suhail (@Suhail) — Wed, 15 Jul 2026 00:18:04 GMT

11/ banging my head against the proverbial hill climbing wall but met a bunch of people who set me in the right direction; SV is very much alive and well with people generous enough to help/share

💬0🔄0❤️3👀2347📊1

⚡ Powered by xgo.ing

If your robot is not absolutely beat up from dogfooded usage, it's probably fake.

Suhail (@Suhail) — Tue, 14 Jul 2026 21:52:38 GMT

If your robot is not absolutely *beat up* from dogfooded usage, it's probably fake.

💬12🔄6❤️152👀24515📊27

⚡ Powered by xgo.ing

You guys are way too focused on Brian Chesky. Build your god damn company.

Suhail (@Suhail) — Tue, 14 Jul 2026 19:31:39 GMT

You guys are way too focused on Brian Chesky. Build your god damn company.

💬1🔄0❤️0👀262📊1

⚡ Powered by xgo.ing

Dom is byfar the most innovative consumer thinker we have in AI. This is like some Twitch meets Vibe...

Suhail (@Suhail) — Tue, 14 Jul 2026 18:59:19 GMT

Dom is byfar the most innovative consumer thinker we have in AI. This is like some Twitch meets Vibe coded app/game thing?

dom hofmann@dhof

excited to share a new toy called bash. it’s a multiplayer coding agent with social built around it

starting with invites and a waitlist, and we’ll open it up as quickly as possible

bash.tv

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬5🔄7❤️86👀30239📊17

⚡ Powered by xgo.ing

.

Suhail (@Suhail) — Tue, 14 Jul 2026 15:28:00 GMT

.

💬13🔄0❤️24👀10978📊13

⚡ Powered by xgo.ing

Steadily knowledge workers are becoming Instruction Workers

Suhail (@Suhail) — Tue, 14 Jul 2026 01:40:48 GMT

Steadily knowledge workers are becoming Instruction Workers

💬23🔄3❤️89👀16085📊29

⚡ Powered by xgo.ing

10/ 64 B300s acquired - if you search hard enough, you'll find what you need

Suhail (@Suhail) — Sat, 11 Jul 2026 14:55:35 GMT

10/ 64 B300s acquired - if you search hard enough, you'll find what you need

💬2🔄0❤️4👀4749📊2

⚡ Powered by xgo.ing

🎯

Suhail (@Suhail) — Fri, 10 Jul 2026 19:45:36 GMT

🎯

The Techno Optimist@technooptimist1

Chamath Palihapitiya @chamath "Bro, there is no failure."

"That is what other people think of you to keep you down. There's no failure. If you were left on an island and it happened, you would just brush it off and move on. It's everybody else that you think is judging you. But then, here's the secret. They don't give a fuck about you. They are living their own lives. So it's your perception of what they think of you."

"There is no failure. There's do and learn. Do and learn. Do and learn. Do and learn."

@StanfordAILab

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬3🔄6❤️134👀35670📊20

⚡ Powered by xgo.ing

9/ made the first hire ❤️ Looking for #2: post training (RLVR/OPSD/etc) or low level model optimiza...

Suhail (@Suhail) — Fri, 10 Jul 2026 15:03:25 GMT

9/ made the first hire ❤️

Looking for #2: post training (RLVR/OPSD/etc) or low level model optimization

💬2🔄0❤️5👀4332📊2

⚡ Powered by xgo.ing

Looking for 64 B200s with IB. DM.

Suhail (@Suhail) — Fri, 10 Jul 2026 13:50:58 GMT

Looking for 64 B200s with IB. DM.

💬4🔄0❤️27👀20225📊7

⚡ Powered by xgo.ing

If anyone has gotten OPSD to work stably, could I ask you some questions?

Suhail (@Suhail) — Fri, 10 Jul 2026 03:05:49 GMT

If anyone has gotten OPSD to work stably, could I ask you some questions?

💬7🔄0❤️18👀13419📊8

⚡ Powered by xgo.ing

To be a bit more generous: I can't tell but perhaps muse-spark-1.1 needs a better harness and/or sea...

Suhail (@Suhail) — Fri, 10 Jul 2026 00:58:15 GMT

To be a bit more generous: I can't tell but perhaps muse-spark-1.1 needs a better harness and/or search capability given the benchmarks. Whatever they have on the dev.* website is not helping them.

💬5🔄0❤️63👀14967📊10

⚡ Powered by xgo.ing

My evaluation of some of the models over past ~24 hours testing/building: GPT 5.6 Sol is excellent....

Suhail (@Suhail) — Fri, 10 Jul 2026 00:55:10 GMT

My evaluation of some of the models over past ~24 hours testing/building:

GPT 5.6 Sol is excellent. You can feel the care in it. Bravo as usual OpenAI.  

muse-spark-1.1 is way off the mark after my testing of some real questions. Bad formatting, not thorough, doesn't dig deeper, seemingly afraid to respond at length, too utilitarian. Something is off.  

Cursor Grok 4.5 has been joyful to use. I ask it lots of side questions while Claude codes. The fast mode is a real delight each day. It really understands, chomps through the code base, and has gotten me to trust it more each day.

💬53🔄43❤️1339👀179263📊196

⚡ Powered by xgo.ing

That you can ask AI "why, why, why" as many times as you want like a 5-year old is a modern marvel o...

Suhail (@Suhail) — Thu, 09 Jul 2026 23:02:48 GMT

That you can ask AI "why, why, why" as many times as you want like a 5-year old is a modern marvel of this century.

💬32🔄35❤️553👀162646📊89

⚡ Powered by xgo.ing

Looking for 64 B200s. DM if you know a vendor.

Suhail (@Suhail) — Tue, 07 Jul 2026 13:48:12 GMT

Looking for 64 B200s. DM if you know a vendor.

💬8🔄2❤️23👀19881📊10

⚡ Powered by xgo.ing

They made a website powered by nuclear: https://t.co/NquIry2l0U 🤯

Suhail (@Suhail) — Wed, 01 Jul 2026 23:54:24 GMT

They made a website powered by nuclear: nuclearwebsite.com 🤯

💬1🔄1❤️10👀3903📊3

⚡ Powered by xgo.ing

Uh, huuuuuuuuuuuuge.

Suhail (@Suhail) — Wed, 01 Jul 2026 23:38:20 GMT

Uh, huuuuuuuuuuuuge.

Atoms Not Bits@AtomsNotBits

BREAKING: @valaratomics is partnering with @nvidia to explore the development of a 30 MW nuclear-powered AI data center in Emery County, Utah, pairing microreactor energy with next-generation cooling designed for near-zero water consumption.

🔗 View Quoted Tweet

💬8🔄10❤️225👀43780📊25

⚡ Powered by xgo.ing

Cost per task is going to be a meaningful metric these next 12 months.

Suhail (@Suhail) — Wed, 01 Jul 2026 21:14:20 GMT

Cost per task is going to be a meaningful metric these next 12 months.

Alex Atallah@alexatallah

On price/token != cost/task:

To help internalize this fact for yourself, run Terminal-Bench using Haiku and then using Opus.

Here are the results for a 15-task subset. Haiku is 10x the cost!

🔗 View Quoted Tweet

💬8🔄7❤️46👀14470📊12

⚡ Powered by xgo.ing

Some very cool people in the DMs. Thank you - will get to you soon. Caught a cold.

Suhail (@Suhail) — Tue, 30 Jun 2026 22:06:10 GMT

Some very cool people in the DMs. Thank you - will get to you soon. Caught a cold.

💬2🔄0❤️6👀5425📊3

⚡ Powered by xgo.ing

Looking for an early hire in SF for helping with model optimization: spec decoding, GPU kernels, poo...

Suhail (@Suhail) — Tue, 30 Jun 2026 15:37:46 GMT

Looking for an early hire in SF for helping with model optimization: spec decoding, GPU kernels, pooling infra, and wants to go deep on vLLM/sglang. Hunger/interest over experience in this case. If you know anyone, DMs open. They’d work directly with me.

💬30🔄12❤️281👀48782📊70

Suhail(@Suhail)

There is going to be 10x more distillation research with the threat of open weight model restriction...

Correct. You want safe models? This is what you do. Tons of software we use everyday is made by peo...

Glad to see Elon's position on this. I really think highly of him coming out on it even though it do...

It sounds like lobbying is going well!

Now I can go to bed. Good luck little AI research agent. May your wandb metrics avoid collapse.

If you want to help adjust the narrative / direction of open weight models, use of Chinese OSS, avoi...

Anthropic has figured out the main game is making everyone operate in a state of fear. It's a consta...

The main thing I've learned this week is that Anthropic's lobbyists are world-class. They are absolu...

If Chinese AI gets added to the entity list it's going to be hard for any American corporation to us...

It brings me tremendous peace to know Emil is completely aware of the regulatory capture going on he...

They *distilled* humanity. The hypocrisy is unreal. The best gift is to give it away, at the lowest ...

The Chinese do not need to distill anymore. They've climbed up the proverbially tech tree that to cl...

Product quality always seems to go down hill when you start thinking to yourself “You know, we sure ...

Literally frog boiling the app layer. There needs to be a neutral party in this.

A company will announce a blog post and some money and people will be like "Ok, but how do you compe...

If your core technology continues to get commoditized (i.e. Kimi 3), you must go up the stack. The l...

It's pretty cool that this technology that you have to have $10B to make is just freely available to...

I see a lot of “without a doubt, they do” and not a lot of evidence or prove of life on methods. You...

Well deserved. Valar is going to be huge.

Talked to some great friends today and completely lost track of time. I have a rule these days where...

Every single credible researcher I've talked to these past few weeks has said the "distillation" fro...

Interestingly, you can almost predict what's coming next based on the timeline. As my son would say:...

I was curious about the last 4 of months of AI releases: Opus 4.7 · 4/16 Kimi K2.6 · 4/20 GPT-5.5 ·...

The ability for AI to visualize things for me has been very helpful. Previously it'd be far too much...

Jeez, who isn't going into the make-RL-env business???

The vertical integration of post-training infra as a service has truly begun.

If you sell synthetic data, real user prompts that people use in AI tools, analyzed user trajectorie...

Must relocate to SF.

I just realized we passed the 1 year anniversary of Soham Parekh incident. Things have changed so mu...

Perplexity is very good at building these benchmarks. Whenever I study them in-depth, it’s always th...

11/ banging my head against the proverbial hill climbing wall but met a bunch of people who set me i...

If your robot is not absolutely *beat up* from dogfooded usage, it's probably fake.

You guys are way too focused on Brian Chesky. Build your god damn company.

Dom is byfar the most innovative consumer thinker we have in AI. This is like some Twitch meets Vibe...

.

Steadily knowledge workers are becoming Instruction Workers

10/ 64 B300s acquired - if you search hard enough, you'll find what you need

🎯

9/ made the first hire ❤️ Looking for #2: post training (RLVR/OPSD/etc) or low level model optimiza...

Looking for 64 B200s with IB. DM.

If anyone has gotten OPSD to work stably, could I ask you some questions?

To be a bit more generous: I can't tell but perhaps muse-spark-1.1 needs a better harness and/or sea...

My evaluation of some of the models over past ~24 hours testing/building: GPT 5.6 Sol is excellent....

That you can ask AI "why, why, why" as many times as you want like a 5-year old is a modern marvel o...

Looking for 64 B200s. DM if you know a vendor.

They made a website powered by nuclear: https://t.co/NquIry2l0U 🤯

Uh, huuuuuuuuuuuuge.

Cost per task is going to be a meaningful metric these next 12 months.

Some very cool people in the DMs. Thank you - will get to you soon. Caught a cold.

Looking for an early hire in SF for helping with model optimization: spec decoding, GPU kernels, poo...

They distilled humanity. The hypocrisy is unreal. The best gift is to give it away, at the lowest ...

If your robot is not absolutely beat up from dogfooded usage, it's probably fake.